A 3-D anthropometric-muscle-based active appearance model

This paper describes a novel method for modeling the shape and appearance of human faces in three dimensions using a constrained three-dimensional (3-D) active appearance model (AAM). Our algorithm is an extension of the classical two-dimensional (2-D) AAM. The method uses a generic 3-D wireframe model of the face, based on two sets of controls: anatomically motivated muscle actuators to model facial expressions and statistically based anthropometrical controls to model different facial-types. The 3-D anthropometric-muscle-based model (AMBM) of the face allows representing a facial image in terms of a controlled model-parameter set, hence, providing a natural and constrained basis for face segmentation and analysis. The generated face models are consequently simpler and less memory intensive compared to the classical appearance-based models. The proposed method allows for accurate fitting results by constraining solutions to be valid instances of a face model. Extensive image-segmentation experiments have demonstrated the accuracy of the proposed algorithm against the classical AAM.